1. Piecewise-deterministic Quasi-static Pose Graph SLAM in Unstructured Dynamic Environments.
- Author
-
Deeb, Amy, Seto, Mae, and Pan, Ya-Jun
- Abstract
This research is inspired by the challenge of providing situational awareness to a ship looking for leads in the ice to navigate the Canadian Arctic. To address this challenge, this paper proposes piecewise-deterministic quasi-static simultaneous localization and mapping (PDQS-SLAM). PDQS-SLAM relaxes the static assumption that is inherent to traditional SLAM algorithms by incorporating landmark kinematic motion models in the graph construction. It then generalizes to more complex dynamic environments by detecting inevitable disruptions to the landmark’s motion model. This is achieved by augmenting the definition of a loop closure factor to include a state, governed by a finite state machine, that captures the edge’s behaviour over time. Simulations and laboratory experiments demonstrate that the localization and mapping errors of PDQS-SLAM in a dynamic environment with mobile landmarks that experience changes in their motion model were similar to the baseline static SLAM in a static environment. This represents a capability that is needed for persistent autonomous operations in complex environments populated with slowly-evolving landmarks, such as the Canadian marine Arctic. [ABSTRACT FROM AUTHOR]
- Published
- 2022
- Full Text
- View/download PDF